the behavioral genetics of psychopathology a clinical guide - kerry l. jang

Behavioral geneticists are just as concerned withthe influence of the environment and its interplay with genetic factors.. Genetic effects refer to the influence genes have on thedevelop

The Behavioral Genetics

of Psychopathology

A Clinical Guide

This page intentionally left blank

The Behavioral Genetics

of Psychopathology

A Clinical Guide

Kerry L Jang

University of British Columbia

LAWRENCE ERLBAUM ASSOCIATES, PUBLISHERS

2005 Mahwah, New Jersey London

Copyright © 2005 by Lawrence Erlbaum Associates, Inc.

All rights reserved No part of this book may be reproduced in any form, by photostat, microform, retrieval system, or any other means, without prior written permission of the publisher Lawrence Erlbaum Associates, Inc., Publishers

10 Industrial Avenue

Mahwah, New Jersey 07430

www.erlbaum.com

Cover design by Kathryn Houghtaling Lacey

Library of Congress Cataloging-in-Publication Data

Jang, Kerry L (Kerry Leslie),

1962-The behavioral genetics of psychopathology : a clinical guide / Kerry L Jang,

p cm:

Includes bibliographical references and index.

ISBN 0-8058-4326-4 (alk paper)

1 0 9 8 7 6 5 4 3 2 1

Chapter 2 The ABCs of Behavioral Genetics

Chapter 3 Classification and Diagnosis

PART II: THE BEHAVIORAL GENETICS

OF THE COMMON MENTAL DISORDERS

Chapter 4 The Mood Disorders

Chapter 5 The Personality Disorders

Chapter 6 The Anxiety Disorders

Chapter 7 Substance Use Problems

viiix

31544

6787111132

vi CONTENTSChapter 8 Schizophrenia and the Psychotic Disorders 150Chapter 9 Recapitulation 165Further Reading 172References 174Author Index 195Subject Index 205

Many of the ideas expressed in this book have been shaped by the work of manyfriends and collaborators over the years, particularly John Livesley, Steven Taylor,and Murray Stein, and I thank them all for sharing I am sure they will tell me that Ihave misrepresented or misunderstood them after reading this book, but that willgive us another excuse to meet, drink beer, and talk about it

Special thanks go to Roseann Larstone, my long-suffering assistant who hashelped me from the outset of this project to keep track of the many drafts that I reg-ularly misplaced as well as to edit, type, and retype the ones I found Her languageduring this time was of a form and style that would make a longshoreman proud.Many thanks to Dr Bonnie Wiese, who bravely read the first draft and provided in-valuable comments on what busy clinicians might care to read about, which helpedshape the content of the book

Finally, a very large thank you to all the career-preparation students fromWindermere, Sir David Thompson, John Oliver, Lord Templeton, Sir CharlesTupper, and Lord Byng secondary schools (Vancouver School District) for theirdiligence in checking and correcting references I hope the experience has not de-terred them from a career in research

This book was supported, in part, by a New Emerging team Grant (PTS-63186)from the Canadian Institute of Health Research, Institute of Neurosciences, MentalHealth and Addiction

vii

This page intentionally left blank

Nurture depends on genes and genes need nurture.

—Ridley (2003)

This book explores the theories, etiology, measurement, diagnosis, and treatment

of psychopathology from the perspective of behavioral genetics, a field of enquiry

broadly concerned with the inheritance of emotional and behavioral patterns.Why do clinicians need to know about the new findings being reported every day

by behavioral geneticists? The aim of treatment is always to change emotional andbehavioral patterns An understanding of the genetic influences that contribute tobehavioral variability and change helps practitioners and patients plan realistic

goals and develop effective strategies to reach them For many people, the term

be-havioral genetics conjures up images of busy automated laboratories searching for

susceptibility genes, a job known as genomics With the mapping of the human

ge-nome, there is no doubt that genomics will continue to be a large part of what havioral geneticists do However, identifying the susceptibility genes is only oneeffort The real impact of behavioral genetics lies with studies that estimate the ef-fect of identified or hitherto unidentified genes on behavior, a task that has been la-

be-beled behavioral genomics (Plomin & Crabbe, 2000) This book addresses the

impact of behavioral genomics on how psychopathology is conceptualized and proached in our daily work

ap-This book is not just about genetics In February 2001 it was announced that thehuman genome contains 30,000 genes, rather than the 100,000 originally expected.This startling revision led some to conclude that there are simply not enough genes

to account for all the different ways people behave and that behavior must also be

ix

shaped by environmental factors Behavioral geneticists are just as concerned withthe influence of the environment and its interplay with genetic factors This bookwill try to spend as much time examining the role of experience as biological factors

on the development of mental illness

Inevitably, books on genetics have to log recent findings in the field ever, it is not my intent to simply list results outlining what is genetic and what isnot, but rather to try to answer the question, "What does knowing the relativeinfluence of genes and the environment mean at a psychological level of analy-sis?" What does this mean? This is best answered with an example It has been ar-

How-gued that diagnostic systems like the Diagnostic and Statistical Manual of Mental

Disorders IV (American Psychiatric Association, 1994) produce reliable

diagno-ses of questionable validity, because the system depends on symptom counts tomake a diagnosis with little or no mention of the underlying causes of disease(Helmuth, 2003) Behavioral genetics sheds light on these causes, and with aclearer vision of causes we can investigate the validity of current diagnoses, ordevelop new diagnostic entities based on the degree to which symptoms share acommon genetic basis

This book is divided into nine chapters The first chapter delineates what ioral genetics is and what it is not, and the criticisms leveled at the field This chapteralso introduces some ideas on how a biologically based genetic psychopathology isapproached with psychotherapy Chapter 2 is devoted to methods It describes inbasic terms how genetic and environmental effects are estimated and the principalmethods used to identify susceptibility genes Chapter 3 is devoted to diagnosis.This chapter shows how behavioral genetic research challenges the fundamentalideas underlying current nosological and classificatory systems and describes somedirections for how they might be revised in the future

behav-Chapters 4 through 8 review the most common classes of adult pathology Each is a selective survey of the published research in the past decade thatprovides a sense of what has been studied (as not all forms of psychopathology havecome under the behavioral genetic microscope) and how consistent the findingshave been Each highlights one theme that is important in an evolving comprehen-sive theory of the linkages among genetic influences, environmental factors, andpsychopathology

psycho-Chapter 4, on mood disorders, introduces the idea of differential heritability—that psychopathology does not necessarily exist as a monolithic and genetically ho-mogeneous entity, but rather that each symptom is influenced to differing degrees

by a multitude of genetic and environmental factors Chapter 5, on the personalitydisorders, presents the dimensional model of psychopathology in which illness isconceptualized as the extremes of normal function, and illustrates how behavioralgenetics has been used to test the validity of this model

Of all the psychopathologies discussed in this book, the anxiety disorders havebeen shown to be influenced by the greatest variety of nonheritable effects Chapter

PREFACE xi

6 demonstrates the importance of the environment, specifically, the mechanisms

by which the environment (e.g., learning) influence the development of disorder.Chapter 7, on substance use, examines the direct and indirect roles of genetic andenvironmental effects in alcohol, tobacco, nicotine, and illicit drug use and the rela-tionship between these substances to explain polysubstance use Chapter 8, on psy-chotic disorders, traces the ways in which behavioral genetic research has providedsupport for the neurodevelopmental model of schizophrenia

The final chapter of this book recapitulates the main findings and also draws tention to new threats to research efforts I hope that this book helps all readers tomake sense of behavioral genetics, and to integrate genetical thinking into dailywork, and that it provides a much more informed perspective on mental disorders

at-This page intentionally left blank

I Behavioral Genetic Basics

This page intentionally left blank

—Uhl and Grow (2004, p 223)

At first glance, research in behavioral genetics appears to be irrelevant to clinicalpractice Rarely do articles comparing the efficacy of different psychotherapeuticapproaches make reference to findings in genetics By the same token, few behav-ioral genetic research reports discuss the clinical implications of their findings.The independence of science and practice was noted by McClearn, Plomin,Gora-Maslak, and Crabbe (1991), who wrote that, despite the fact that many be-havioral geneticists have a background in the social sciences, genetic perspectives

on behavior have " not yet completely woven into the pattern of psychologicaltheory " (p 222)

An important purpose of this book is to begin the integration of genetics intoclinical thinking and research It is often thought that the only clinical application

of genetic research is the development of drug therapies to counteract the offendinggene's product The first major step in this process is to determine if the genes hy-pothesized to be associated with a disorder are actually present in patients with thatdisorder Once a gene is linked to a specific disorder, the biochemistry associatedwith the gene becomes the focus to determine the intracellular mechanisms by

3

4 1 INTRODUCTION

which abnormal behavior is produced Many researchers in behavioral genetics aretrained in medical genetics and other medical specialties Their approach is to workfrom the bottom up: They take the fundamental unit of analysis to be the gene andits variants Frustratingly, successes have been few and far between

All is not lost, because a great deal of what behavioral geneticists do is to study theeffects of these as-yet-to-be identified genes They take a top down approach that be-gins with recognized disorders (e.g., the symptoms and signs of mental illness) anduses genetically informative samples, such as twins or adoptees, to determine if indi-vidual differences in the disorder are due to genetic variations or to changes in envi-ronmental conditions Genetic effects refer to the influence genes have on thedevelopment of individual differences in behavior relative to the influences of learn-ing, experience, and environmental conditions The size of genetic and environmen-tal effects can be estimated for a single disorder such as major depression or forindividual symptoms like sadness or insomnia We can estimate the relative geneticand environmental impact on virtually any behavior that can be measured reliably.This introductory chapter will explain some important basic concepts, such as thedefinition of illness, outline some of the criticisms leveled at behavioral genetic re-search, and finally describe some psychotherapeutic approaches being developed toaddress behavior whose expression may be fixed by inherited factors

pressed as percentages For example, h 2 - 40% means that 40% of the differences

observed between people are directly attributable to genetic differences between

them A popular method to estimate h 2 is to measure the similarities (e.g., usingthe simple correlation coefficient) of infants who were adopted and raised by bio-logically unrelated families to their biological family members after they havereached adulthood Any similarities between the adopted children and their bio-

logical relatives can only be due to the genes they share, yielding an estimate of h 2

It follows that if h 2 = 40%, then 60% of the differences between people must be due

to environmental factors, including the influence of family environment Family

environmental effects, symbolized as c 2 , can be estimated by comparing the

simi-larity of adopted children to their adoptive families Because they have no genes incommon, any similarity can only be due to the fact that they all share the samehome environment

Few disorders are entirely under genetic control Even if a disorder were 100%heritable, the expression of the relevant genes might still be controlled by environ-mental factors such as learning experiences or exposure to a specific environmentalcondition There are many examples of this phenomenon in the medical literature

An oft-cited example is phenylketonuria (PKU), a form of mental retardation

When R Adron Harris and his team at the University of Texas, Austin, screened 10,000 genes in the frontal and motor cortexes of alcoholics, they found changes in the

expression of 191, they reported in the Journal of Neurochemistry.

Alcohol seems to cause a "selective reprogramming" of brain genes in areas involved

in judgement and decision making, says Dr Harris Among them: genes that code for myelin, whose loss may impair cognition and judgement.

Antidepressants may also alter genes The conventional wisdom is that drugs such as Prozac work by blocking re-uptake by brain neurons of the neurotransmitter serotonin But Prozac starts doing that in 24 hours Why, then, do such drugs typically take weeks

to lift depression? "The hunch is that Prozac work by altering gene expression, maybe be causing sprouting of new neurons and remodeling of synapses," Dr Harris says Experience, too, can affect gene expression How much a mother rat handles and licks off her offspring—an environmental influence if ever there was one—has an astonish- ing effect: It determines whether genes that code for receptors for stress hormones in the brain are expressed or not And the level of those receptors affects how a rat reacts

to stress Rats with attentive moms were much less fearful and more curious, finds chael Meaney of McGill University in Montreal Rats that got less maternal handling grew up to be timid and withdrawn in novel situations.

Mi-Rats are not long-tailed people, so you can't infer that maternal affection affects gene expression and thus temperament in babies, too But something sure does There is no shortage of evidence that intelligence, shyness, impulsivity, risk-taking, and illnesses have a genetic component.

But identical twins, who have the same genes, don't have identical traits: One twin might

be schizophrenic and the other not, one might be shy and the other outgoing, one might get a "gene-based" cancer and the other not The difference between identical twins is the experiences they have and, if I may speculate, which of their genes are expressed.

What signal from the environment keeps schizophrenia-related genes silent? What activates IQ-lifting genes? Whatever it is, even a short-lived environmental signal might turn on genes that tell neurons how, and how much to grow That would leave

an enduring mark: Neutral circuits would be complex or simple, and different brain regions would be strongly linked or not From such neuronal differences arise differ- ences in intelligence and personality, health, and temperament.

Linking specific environmental influences to gene activity would have been a pipe dream only a few years ago But the new technology of microarray analysis, in which

"gene chips" reveal which DNA in a sample of tissue is expressed and which quiescent,

is making such discoveries possible.

This past April, in one of their coolest uses so far, gene chips showed that the difference tween human brains and chimp brains is not which genes each brain has Those are nearly identical The difference is which genes are turned on and which are switched off Ironically, the recognition that genes depend on the environment follows hard on the heels of genetics' greatest triumph: sequencing the human genome But what's now clear

be-is that the more we learn about genetics, the more we'll see that genes are not destiny.

FIG 1.1 Gene-environment interplay From "Even Thoughts can Turn Genes 'on'

and 'off,'" by S Begley, June 21, 2002, San Francisco Chronicle Copyright 2003 by the

Associated Press Reprinted with permission from Dow Jones & Company, Inc.

6 1 INTRODUCTION

caused by an excess of the amino acid phenylananine PKU is controlled by the fect of a single gene found on chromosome 12 Inheriting the gene for PKU does notnecessarily mean that mental retardation is inevitable Phenylananine is present inmany foods and simply eliminating these foods from the diet of a PKU-gene-carry-ing infant will prevent the development of genetically mediated mental retardation.This is an example of gene-environment interplay, examples of which can be foundfor all kinds of human behavior (see Fig 1.1) Behavioral geneticists have long beenawed by the power of the environment and spend as much time and effort identify-ing these effects as they do in the search for genes and estimating genetic effects

ef-The Unifactorial Model of Disease

The general public's conception of what constitutes a genetic disease is looselybased on a model of a single gene of major effect—if a person has this gene, they willdevelop the disease However, Temple, McLeod, Gallinger, and Wright (2001) de-scribed how naive this popular conception is:

Human genome sequencing will reveal thousands of genetic variations among individuals that many will assume are associated with disease But translating such genotypic differences (genetic characteristics) into pheno- typic states (visible characteristics) is prone to pitfalls For example, genetic abnormalities differ in their penetrance (that is, not everyone carrying a ge- netic abnormality will suffer from adverse consequences); environmental ef- fects have not been taken into consideration; and many diseases have complex etiologies that depend on a number of different genes There are very few diseases that are caused by a single gene mutation Automated genomic sequencing is becoming increasingly sophisticated, but distinguish- ing between normal variations in genes (polymorphisms) and alterations that are detrimental (mutations) remains extremely difficult This difficulty will have direct consequences for genetic counselors, who must advise indi- viduals about the presence of genetic abnormalities, what they mean, and which treatment or prophylaxis to follow, (pp 807-808).

The Threshold Liability Model of Disease

How geneticists understand disease is not much different from how behavioral entists presently conceptualize disorder This is the classic threshold liability modelthat assumes that behavior is normally distributed in a population (see Fig 1.2).The severity of behavior, such as that measured by a self-report or clinical ratingscale, is plotted along the horizontal or x-axis Plotted along the vertical or y-axis isthe number of people displaying behavior at a given level of severity The distribu-tion of scores in this hypothetical population is split into areas representing threedistinct groups The vast majority of people in this population fall within Area A,the normal range of expression Area B, spectrum conditions, represents the pro-portion of the population whose behavior does not meet the full criteria to be con-

sci-GENETIC EFFECTS

FIG 1.2 Threshold liability model of disease.

sidered clinically abnormal (Area C), but is not quite typical enough to beconsidered normal, that is, fitting the criteria for inclusion in Area A

The breadth of Areas A, B, and C is defined by the clinician or researcher andneed not be symmetrical around the arithmetical average For example, the point orthreshold that differentiates normal from abnormal behavior can be based on sta-tistical criteria (e.g., any score that falls two deviations above or below the mean) or

it can be defined as a cut-off on a questionnaire or clinical screen that reliably entiates patients from healthy controls For example, clinical research on the Anxi-ety Sensitivity Index (ASI, Peterson & Reiss, 1992) showed that any score above 25 isclinically significant, irrespective of how the scores are distributed in a population

differ-It is important to remember that, depending on how disorder is defined, it is ble to delineate fewer or more areas (e.g., affected, unaffected vs unaffected, slightlyaffected, moderately affected, definitely affected, severely affected) to describe thedistribution of disorder in a sample or population

possi-The threshold liability model is also central in behavioral genetics However,behavioral geneticists tie severity along the x-axis to the relative influence of ge-netic and environmental factors that determines the number of people falling intoeach of the areas This book presents several ways genetic and environmental in-fluences are thought to increase and decrease the variability of the distribution

8 1 INTRODUCTION

For example, individuals in Area C inherited the specific gene forms for mental ness, had been exposed to the required traumatic events, or were exposed to mul-tiple events over time to cause illness In contrast, people in Area A did not inheritthe genes or may not have been exposed to the requisite traumatic events; people

ill-in Area B may have ill-inherited some but not all of the genes for the disorder or havenot experienced the critical number or kinds of traumatic events Alternatively,every person may have inherited the genes that determine susceptibility to disor-der, but they remain silent unless triggered into action by exposure(s) to specificenvironmental stressors or remain silent forever in environments that suppresstheir actions indefinitely Similarly, people in Areas A and B may have inheritedgenes that protect them from the effects of trauma that people in Area C did notinherit The point is that there are multiple ways genetic and environmental influ-ences work together to produce behavior, that behavior is multifactorial, caused

by the action of several genes and experiences

Other factors that influence the distribution of behavior include the study tion (e.g., patient, normal control, general population samples), the sensitivity andspecificity of the behavioral measures (e.g., measures that screen for disorders vs.those that assess specific symptoms), and the content of the measure (e.g., designed toassess normal or abnormal behavior) For example, if the study population consisted

popula-of unaffected general population study subjects who completed scales that assessedextreme behavior, then there would be no people in Area C, with all falling in Areas Aand B At best, a study based on this population would not be estimating theheritability of a disorder per se, but the heritability of a spectrum condition

CRITICISMS OF BEHAVIORAL GENETICS

Over the years, a number of objections to behavioral genetic research have beenraised, which have no doubt contributed to its apparent lack of impact in clinicalcircles Concerns range from questions about the adequacy of research design to thesociopolitical implications of findings In this section, I offer a brief review of thecriticisms and responses, both philosophical and empirical

Are Heritability Estimates Uninformative?

It is not apparent what scientific purposes are served by the sustained flow ofheritability numbers for psychological characteristics Perhaps molecular geneticistsneed those numbers to guide their search for underlying genes? Perhaps clinical psy-chologists need those numbers to guide their selection of therapies that work? Or per-haps educators need those numbers to guide their choice of teaching interventionsthat will be successful? We have seen no indication of the usefulness of heritabilitynumbers for any of these purposes (Kamin & Goldberger, 2002, p 93)

Is this a valid criticism? It has been argued that the heritability statistic is andhas been useful for the very reasons Kamin and Goldberger (2002) suggest.Boomsma, Martin, and Machin (1997) wrote that the raison d'etre of twin re-

CRITICISMS OF BEHAVIORAL GENETICS 9

search was to identify behaviors that would be suitable for genotyping analysis cause there is no sense in looking for a susceptibility gene if the disorder is notshown to be heritable in the first place Moreover, simply demonstrating that aspecific behavior is heritable can be paradigm shifting For example, finding thatsocial attitudes have a substantial heritability challenges the common assumptionthat they are acquired entirely via social learning (see Olson, Vernon, Harris, &

be-Jang, 2001) Most importantly, it must be remembered that h 2 is just the startingpoint Estimating genetic effects on a single variable is the first step to estimatingthe heritability of a second variable that leads to estimating whether these geneticeffects are common to both variables, and this information can be used to explainwhy behaviors coaggregate the way they do

Are Behavioral Genetic Methods Inherently Flawed?

One of the classic ways heritability is estimated is by comparing similarities ofmonozygotic (identical) twins who were separated at birth and raised in differentfamilies This comparison provides a direct estimate of genetic effect because thesetwins share 100% of their genes but grew up in different environments and any sim-ilarities can only be due to their genetic similarity, thus yielding a direct estimate of

h 2 Kamin and Goldberger (2002) questioned the validity of heritability estimates

based on this method because of concerns over:

1 The representativeness of samples Are the results of studies of twins

generalizable to the general population? Do twins live in unique stances and receive special treatment from others because they are twins?

circum-2 The accuracy of the data This includes issues about the reliability and

va-lidity of self-report versus observer reports and the extent of contact tween separated twins They noted that most behavioral genetic studiesrely on self-report questionnaires and often the responses to only a fewitems from a scale

be-3 Measurement of selective placement effects For example, the similarity of

twins would be spuriously inflated if twins were placed in homes of familiesthat were genetically similar (e.g., placement based on ethnicity)

It would be fair to characterize these concerns as very much yesterday's news havioral geneticists have addressed them and a large body of relevant research wassummarized in Boomsma et al., (1997), an article that, incidentally, predatesKamin and Goldberger's critique by 5 years but was not cited by them However,the fact that the criticisms continue to be repeated indicates that behavioral geneti-cists have not done a very good job of publicizing their efforts

Be-It is also important to point out that some of the criticisms are not unique to havioral genetic research For example, the complaint that many studies rely onself-report measures is one that can be leveled at all kinds of behavioral research Infact, the behavioral genetic research on psychiatric disorders typically uses data col-

be-10 1 INTRODUCTION

lected by clinical interview and other-report (e.g., from parents, roommates,spouses, or teachers) in addition to long and short forms of many popular self-re-port scales The choice of instrument used in any study tends to reflect the training

of the researcher or its recognition as the best way to measure a particularpsychopathology What the reader should look for is a convergence in results acrossstudies regardless of a scale's response format

Are Behavioral Geneticists Eugenicists?

Perhaps some of the resistance to behavioral genetics stems from the field's ent association with the eugenics movement and the spectre of selective breeding ofhumans for desirable traits The association began with the publication of Arthur

appar-Jensen's 1969 article in the Harvard Educational Review, in which he presented early

behavioral genetic research that showed that cognitive ability (IQ) had a large table component These findings led some to reason that, if differences in mentalabilities are inherited and if success requires those abilities and if earnings and pres-tige depend on success, then social standing will be based to some extent on inher-ited differences among people (Bernstein, 1971, p 43)

heri-People quickly rallied against the IQ test and anything associated with it A cal response was: "The intelligence test has been used more or less consciously as aninstrument of oppression against the under-privileged—the poor, the foreign bornand racial minorities" and "a critical review of the literature produces no evidencewhich would convince a prudent man to reject the hypothesis that intelligence testscores have zero heritability" (Kamin, 1974, p 1), setting the stage for the socio-political "nature versus nurture" battle that has been associated with behavioral ge-netics ever since

typi-However, as early as the late 1970s it was clear to many that behavioral geneticswas not associated with either side in this debate For example, Charles Crawford(1979) wrote a tightly argued paper entitled "George Washington, Abraham Lin-coln and Arthur Jensen: Are they compatible?" Crawford demonstrated that takingeither a pronature or pronurture position was irrational and that the emotion thedebate generates stems from the conflict between basic American values and thetruth of scientific research

Crawford (1979) began by describing two core values of American society Thefirst is that truth despite the consequences is an essential element in the ascent of hu-manity, exemplified by George Washington's statement: "Father, I cannot tell a lie,

I did it with my little hatchet." The second is exemplified by Abraham Lincoln's puted) statement that, "If my father's son can become President, so can your fa-ther's son." This is no less than the expression of the American dream, which saysthat, with hard work and determination, anything is possible The conflict arises be-cause one cannot believe in equal access to the American dream and the inheritance

(re-of individual differences in intelligence while maintaining the importance (re-of truth!More importantly, Crawford showed that (see Fig 1.3) "correctly believing in an

Possible state of nature

Belief

Individual differences in intelligence are largely genetic in origin

Individual differences in intelligence are largely environmental in origin Individual differences in

intelligence are largely

Correct decision Possible consequences:

Social Darwinism Socialized medicine, guaranteed Annual income, etc.

Inventing new American Dream

Correct decision Possible consequences:

Cultural imperialism Equal access to American Dream

Orwell's 1984; Huxley's Brave New World

Type I error Possible consequences:

Inappropriate special schools for minority groups Inappropriate marriage, adoption, and miscegenation laws

Waste of intellectual talents of many citizens

' This would consist in incorrectly believing in environmental determination of individual differences in intelligence.

' This would consist in incorrectly believing in genetic determination of individual differences in intelligence.

FIG 1.3 Possible consequences of outcomes of the nature-nurture debate From "George Washington, Abraham Lincoln, and Arthur Jensen:

Are They Compatible," by C B Crawford, 1979, American Psychologist, 34, pp 664-672 Copyright © 1979 by American Psychological

Associa-tion Used with permission.

12 1 INTRODUCTION

environmental outcome does not necessarily lead to Utopia in our grandchildren'sday and correctly believing in a hereditarian outcome does not necessarily lead tosocial Darwinism" (p 664)

Dean Hamer and Peter Copeland came to the same conclusion in their 1998

book Living With Our Genes In this book, they examined what was known about

inheritance of everyday human behaviors, including sex, alcohol and drug use, lence, eating habits, and personality They too found that a polemical debate waspointless and concluded that the "DNA map offers the possibilities and predictionsbut no certainty" (p 308), but cheekily added "free will is alive and well, and proba-bly genetic" (p 314) Behavioral geneticists have never taken the extreme positionthat heritability equals inevitability and they actively research the role of environ-mental factors Among behavioral geneticists, there is simply no debate and the re-search is not driven by ideology!

vio-GENETICAL PSYCHOTHERAPY

Behavioral genetics examines the effect of genes at the level of populations andsamples, but what does a heritable disorder mean at the level of an individual? Atthis level, genetic predispositions likely impose limits on the degree to whichchange is possible and the goals of treatment are to help individuals adapt totheir psychopathology and express it in useful (or at least neutral) ways Wein-berg (1989) put it best: "Genes do not fix behavior Rather, they establish a range

of possible reactions to the range of possible experiences that environments canprovide" (p 101)

Psychotherapeutic approaches tend to focus on psychosocial adversity andemphasize manipulating the environmental conditions, behaviors, orcognitions to effect change Many models of psychotherapy are based on the un-derlying notion that psychopathology is the result of a deficit or conflict In thedeficit model of psychopathology, disorder is characterized by deficits that oc-curred because the early environment failed to provide the necessary ingredi-ents for the child to develop psychologically Change is believed to arise from theprovision of a supportive, empathic, and validating therapeutic environment

In contrast, traditional psychoanalytic approaches exemplify the conflict model

of psychopathology, in which disorder results from defenses against conflicts.The model is attractive because it provides a comprehensive explanation of thedevelopment of disorder and simultaneously offers a coherent therapeutic ap-proach If disorder arises from conflict, it follows that resolution of the conflictusing traditional strategies such as confrontation, clarification, interpretation,and working through conflicts, especially in the transference situation, shouldeffect change

Both models are strongly environmentalist and assert that psychosocial sity is the major contributing factor Genetic effects are acknowledged, but have lit-tle practical impact on a therapist's concepts of psychopathology that would in turninfluence understanding mechanisms of therapeutic change and the goals of ther-

adver-GENETICAL PSYCHOTHERAPY 13

apy How can knowledge about genetic effects be meaningfully integrated into ditional psychotherapeutic practice? Livesley (2001) suggested that conflict anddeficit models need to be supplemented with a vulnerability model of psycho-pathology that explicitly recognizes genetic predispositions It postulates that ge-netic predispositions toward certain illnesses will always exist and that these alsoimpose limits on the degree to which change is possible The underlying principle ofthe model is that the combination of genetic predisposition, environmental adver-sity, and evolving dispositional factors support each other and lead to a system that

tra-is "stably unstable."

As such, the goals of therapy are to modify the level of behavioral expression byeither dampening or amplifying the effects of genetic predispositions and to influ-ence the selection of behaviors through which psychopathology is expressed Forexample, sensation seeking is a highly heritable personality trait However, not ev-eryone with a high score on sensation seeking will express this trait in the same way,nor will different expressions be equally maladaptive Maladaptive expressionsmight include generating excitement by taking an overdose and calling the para-medics or creating public turmoil by threatening to harm oneself or others Moreadaptive expressions might include engaging in high-risk sports or speculating onthe financial markets as a hobby Three basic strategies for managing inheritedpsychopathology are suggested by the model: (a) increasing tolerance and accep-tance, (b) attenuating expression, and (c) progressively substituting more adaptivebehaviors (Livesley, 1999, 2001)

Increasing Tolerance and Acceptance

The basic premise of this strategy is that behavior is relatively fixed and individualsneed to learn to accept their major behavioral characteristics and use them adap-tively Many patients express extreme dissatisfaction and distress about their per-sonal qualities They attack themselves for having certain characteristics, make war

on themselves, in effect Helping patients to understand that their behavior is ited and to recognize that most behavior can be adaptive in some way may reducethis internal conflict Acceptance helps to reduce distress, prevent escalation, andfree the individual to use these fixed behaviors constructively

inher-Livesley (2001) argued that implementing this strategy requires three tasks of thetherapist The first is helping patients understand the core behaviors that definetheir psychopathology and the factors that lead to the development of these charac-teristics The second is helping patients to identify adaptive features of their behav-ior Many common behavioral extremes probably emerged during the course ofevolution because they conferred some kind of adaptive advantage, but in the pres-ent context may no longer do so (e.g., Tooby & Cosmides, 1990) Patients may be-gin to accept their behavior rather than fight it if they recognize its adaptivepotential The therapist can encourage patients to consider the costs and benefits oftheir behavior In the process, maladaptive expressions may be modified and thepatient may begin to recognize ways in which the behavior can be useful Third, re-

14 1 INTRODUCTION

ducing the focus on change and identifying relevant situations and activities allowspatients to use their basic traits as assets Gatz (1990) referred to this as creating orlocalizing a patient's "comfort zone."

Attenuate Expression

The focus in this strategy is teaching patients skills to regulate and control fixedbehavior It assumes that the most malleable component of behavior is the cogni-tive component and that a cognitive-behavioral approach would be the most use-ful way to modify maladaptive cognitive strategies that amplify behavioralexpression Beliefs such as "I can't cope with feelings," "My feelings are always out

of control," or "There's nothing I can do to stop my feelings" can be challengedand reframed With other behaviors it may be possible to teach skills that are com-plementary or incompatible with the behavioral traits, such as stress managementand relaxation training These methods are best introduced gradually (e.g., start-ing with simple relaxation methods) because many patients find more complexrelaxation training aversive

Progressively Substituting More Adaptive Expression

The assumption in this strategy is that the maladaptive behaviors are relatively fixedbut the mode of expression may be changed For example, patients who exhibit ex-treme submissive and dependent behaviors can be taught assertiveness skills Thisbehavioral approach, coupled with efforts to change their beliefs about submission,teaches patients to develop appropriate ways to solicit help and support

The vulnerability model complements traditional approaches to psychotherapy.All therapists can benefit from an appreciation of the limited extent to whichmaladaptive behavior may be eradicated in a person; the goal becomes adaptationand control This can be achieved using standard psychotherapeutic techniquessuch as cognitive restructuring, counseling or insight, and catharsis The vulnera-bility model differs from others only in its explicit recognition of the role of genetics

in setting limits on the degree to which behavior can be changed

SUMMARY

Behavioral genetics research has explored the impact of genes and the environment

on disorder Its results have a direct bearing on our understanding of whatpsychopathology is and in turn influence our understanding of therapeutic changeand the goals of therapy The next chapter reviews the methods used by behavioralgeneticists to estimate genetic and environmental effects and those used to identifygenetic loci underlying psychopathology

The ABCs of Behavioral Genetics

Underlying all behavioral genetic study designs is the principle that relativesshare genes and the additional fact that some relatives, such as siblings, sharemore genes with each other than with their cousins helps us determine whetherheredity is implicated in a disorder This chapter introduces the ways in whichbehavioral geneticists find and estimate the effects of genes and the environ-ment It gives readers sufficiently detailed information on methods and statis-tics used in the field to enable them to digest and critically appraise thebehavioral genetic research reported in the major journals, and to evaluate itsimplications for their patients

FINDING GENES Estimating Genetic Risk

The first step in the search for genes is to identify families in which the risk of veloping the disorder is high compared to families randomly sampled from thegeneral population Risk for a disorder is greater than zero when the frequency ofdiagnosis is greater among genetically related individuals than in a sample ofmatched controls These disorders are also said to be "familial" because they havebeen shown to "run in families."

de-Two kinds of genetic risk are commonly estimated using traditional control family studies: relative risk (RR) and population relative risk (PRR).These estimates are used to identify high-risk families to be included ingene-hunting studies and in genetic counseling that helps patients and their

case-15

2

16 2 THEABCs

families make informed choices about marriage and childbearing RR estimatesthe extent to which a relative of an affected individual is more likely to developthe condition than a relative of a nonaffected individual PRR estimates thechances that the relatives of an affected individual are more likely to be affectedthan the relatives of a nonaffected individual An example of estimating RR andPRR is presented in Figure 2.1 It is important to remember that these risk esti-mates also reflect environmental influences on the disorder Nongenetic factorssuch as home environment, culture, and frequency of contact also contribute tothe similarities between people from the same family The traditional case-con-trol family design cannot separate the influence of shared genes from shared en-vironmental factors

Once the high- and low-risk families have been identified and blood sampleshave been taken from each member, DNA is extracted and the relationship betweenthe DNA and observed behavior is estimated A generic term for any kind of study

that uses DNA analysis is molecular genetic study Molecular genetic studies are

de-signed to tie variations on a specific gene identified in each person's DNA to tions in that person's observable behavior (e.g., presence or absence of the diagnosis

varia-or variation on psychological test scvaria-ores)

Relatives

of schizophrenic Relatives individuals of controls Total

N Relatives with Schizophrenia 9 2 11

N Relatives without Schizophrenia 91 98 189 Total 100 100 200 Computing Risk

Relative Risk (RR) = (9/100) + (2/100) = 4.5

Thus a relative of a schizophrenic patient is 4.5 times more likely to become affected with the disease than the relative of a nonschizophrenic patient.

Population Relative Risk (PRR) = (9/100) + (0.15) = 6.0

For this example, we have assumed that the prevalence of the disorder in the general ulation is 1.5%.

pop-Thus, the relatives of a schizophrenic patient are 6 times more likely to develop the disease than the relatives of a nonschizophrenic patient.

FIG 2.1 Computing relative risk and population relative risk Adapted from ysis of Genetic Data: Methods and Interpretation," by R M Cantor and J I Rotter (Eds.), 1992, The Genetic Basis of Common Diseases (pp 49-70) Oxford: Oxford University Press Adapted with permission.

"Anal-FINDING GENES 17

The Linkage Study

One of the primary molecular genetic methods is linkage analysis Linkage methodsuse the known locations of genes as road signs or markers for the disease gene to ob-tain an approximate idea of where it is located on a chromosome For example, ifthe disease gene is thought to be on a particular chromosome, a known gene on thatchromosome is selected as a marker The marker may or may not be related to thedisease gene Common markers are blood group genes According to Mendel's law

of genetics, the transmission of genes from parent to offspring should be random.Thus, if the disease gene and marker gene were far apart or on different chromo-somes, the probability that they would be passed down together from parent to off-spring is zero Conversely, if the disease gene is physically close to the marker gene,the likelihood that the disease and marker genes would be transmitted together is

high and they are said to be linked.

The term linkage refers to the fact that when genes are in close physical proximity

they literally exchange genetic material during meiosis The exchange of genes fromdifferent chromosomes leads to the production of offspring that have a differentcombination of genes from either parent The probability of two genes, such as the

disease and marker genes, undergoing recombination is called the recombination

fraction, or This fraction varies from zero to 50 The recombination fraction can

be thought of as the distance between the disease and marker genes: A 0 of zero cates close linkage, 30 represents weak linkage, and 50 represents no linkage Insimple terms, 0 indexes the degree to which the disease and marker genes are sharedamong family members of an affected person

indi-Another statistic, called the LOD score (log probability ratio score), is used to

es-timate the actual likelihood that a disease and marker gene will be transmitted gether in a high-risk family The LOD score is computed as:

to-Likelihood of observed pedigree with 0 < 50 LOD = log 10 x

Likelihood of observed pedigree with 9 = 50

The LOD score indexes the probability of any observed linkage between ease gene and marker gene in the family tree of an affected person (thus 0 willhave an actual estimated value of less than 50 in the high-risk family) divided bythe probability that any linkage in the family tree is due to chance (hence, =.50) If the observed probability of linkage equals the probability that the linkage

dis-is due to chance, a LOD score of 1.0 would result, indicating that the level of served linkage is no different from that expected by chance If the observedprobability of linkage is greater than the probability that the linkage is due tochance, the LOD score would be > 1.0 LOD scores of 3.0 are traditionally con-sidered the statistically significant threshold to indicate that the disease andmarker genes are truly linked

ob-18 2 THE ABCs

There are three considerations that set limits on the ability of linkage analysis

to localize the genes for behavior First, the method works best when relatively fewgenes with large effect are implicated in the disorder under study However, it iscurrently thought that behavior is actually multifactorial, or caused by the action

of several genes of small effect Second, linkage requires that the affected status ofeach member of a family can be assigned unequivocally Any misdiagnosis,whether caused by poor assessment methods, uncertainty or lack of specificity indiagnostic criteria, or the presence of a comorbid disorder—even in a single fam-ily member—can have enormous impact on the LOD score Finally, linkage stud-ies require complete data The results are jeopardized if some members of thefamily are unwilling to participate or if data is missing (e.g., affected status of along-dead relative is unknown)

The Association Study

Association studies have become very popular in the search for susceptibility genesbecause they do not rely on family pedigrees Association studies test whether the

hypothesized disease gene (or form of a gene called a polymorphism) is found in

more affected than nonaffected individuals The method got its name because sures of association such as the correlation coefficient are the primary statistics used

mea-to relate genes mea-to behavior A nice feature of association methods is that they arereadily adapted to handle the many different ways behavior is measured in clinicaland research settings For example, by using different correlation coefficients, qual-itative (e.g., presence or absence of illness) or quantitative data can be analyzed withequal ease The strength of the method lies in its use of continuously distributedquantitative data (such as scores on a depression rating scale), where the disorder ismeasured with enough sensitivity to discriminate between a number of levels of se-

verity Association studies using quantitative data are referred to as quantitative

trait loci, or QTL, analyses (see Plomin, DeFries, Craig, & McGuffin, 2003; Plomin

& Caspi, 1998 for a detailed review of QTL methods)

Unlike linkage studies, association studies do not pick road signs to localize theposition of a disease gene on a chromosome Rather, they require that the gene se-lected for analysis actually be involved in the disorder of interest For example, clini-cal studies suggest that the neurotransmitter dopamine is implicated in bipolardisorder and any one of several genes active in dopamine production, transport,

and reception can be selected for analysis These are called candidate genes.

Candidate genes are also selected on theoretical grounds For example, RobertCloninger (1986,1994) developed the "Biosocial Model of Personality" to provideguidance in the selection of candidate genes He hypothesized that there are fourtraits of temperament: harm avoidance, novelty seeking, reward dependence, andpersistence Specific inherited monoamine neurotransmitter systems underlie eachtrait: serotonin for harm avoidance, dopamine for novelty seeking, and norephine-phrine for reward dependence The model has yet to hypothesize a system for per-sistence One of the very first reports of finding a personality gene was a direct result

THE EFFECTS OF GENES AND THE ENVIRONMENT 19

of this model Cloninger, Adolfsson, and Svrakic (1996) reported a significant ciation between different forms of the dopamine DRD4 gene and scores on mea-sures of novelty seeking

asso-The success of association studies greatly depends on identifying a plausiblecandidate gene In the absence of a suitable candidate, a systematic and mechanis-

tic approach called the genome scan can be used to identify possible candidates A

genome scan associates a number of genes from every chromosome (e.g., everygene that sits 10 centimorgans apart1) with disease status The limitation of ge-nome scans is that they require very large sample sizes of affected people and inde-pendent replications to reduce the number of false-positive results one would get

by testing so many genes

As with the linkage study, this method works best when the disease status of ple in the study is not confounded by comorbid psychiatric conditions However,unlike the linkage study, association studies do not require candidate genes withlarge effects Association studies have been very successful in identifying poly-morphisms that have very small effects, accounting for between 3% and 5% of thevariation with acceptable levels of statistical certainty

peo-In summary, linkage and association studies are designed to localize and identifysusceptibility genes for mental illness The identification of genes and their actionwill have major implications for drug development However, the results of thesestudies are compromised when specific methodological requirements are not met,such as quality of diagnosis (e.g., no misdiagnosis or presence of diagnosed orundiagnosed comorbid conditions) or availability of all family members At thistime, these studies can only estimate the effect of one gene at a time and the results

to date suggest that they individually have only a very small effect on behavior

THE EFFECTS OF GENES AND THE ENVIRONMENT

The heritability coefficient introduced in chapter 1, h 2 , is just one of a family of

ge-netic and environmental effects estimated in behavioral gege-netics research Figure2.2 presents a summary of the major effects outlined by Douglas Falconer (1960) in

the classic work Introduction to Quantitative Genetics This equation simply states

that individual differences observed on a measured behavior (often referred to as

the phenotype) is attributable to the sum of the genetic (G) and environmental (E)

differences between people, in addition to differences caused by the interplay of

ge-netic and environmental factors (GE), and error of measurement Measurement

er-ror is assumed to be random and encompasses the vagaries of the clinical "hunch"

to the degree a scale is unreliable

The term G refers to the variability in observed behavior that is attributable to

all sources of genetic influence indexed by h 2B , commonly referred to as broad heritability coefficient The h is for heritability and the subscript B is for broad The

A centimorgan, or cM, is a measure of distance between two genetic loci on a chromosome Two netic loci are 1 cM apart if their probability of recombination is 1%.

ge-20 2 THE ABCs

Behavior X=G + E+ GE+ error Differences in measured behavior X between people are caused by:

Behavior X = Genetic Differences Between People

+ Environmental Differences Between People

+ The Interplay of Genes and the Environment + Errors in Measuring Behavior X

FIG 2.2 Quantitative genetic theory.

h is squared as a consequence of its computation (described later), but also as a

re-minder that the quantity is a variance In most statistical textbooks this is

symbol-ized as a squared term such as 2 when describing a population, or s2 whenreferring to a sample drawn from a population. 2 Heritability estimates index thepercentage of the total variation observed on behavior based on all forms of ge-netic differences between people

h 2B subsumes three types of genetic effect: additive genetic effects (h 2A ), genetic

dominance effects (h 2d , and genetic epistasis (h 2i ) Additive genetic effects are those

that are passed down directly from parent to offspring Genetic dominance andepistasis are called "nonadditive" because their effects are not direct but are ratherdue to the interaction of genes Genetic dominance is variation attributable to theinteraction of genes that occupy the same loci on different chromosomes An exam-ple of genetic dominance effects is the color of a child's hair that is somewhere be-tween the two parents' Genetic epistasis is caused by the interaction of genes from

different loci Estimates of h 2i have been difficult to detect because they have beenconsidered small (e.g., Plomin, DeFries, & McClearn, 2000)

How to Estimate Heritability

The first step in measuring any genetic or environmental effect is to measure the

similarity of relatives Similarity is measured by the correlation coefficient, r

Sev-eral types of correlation coefficients are used in behavioral genetic research, the typebeing dictated by several factors including the level of measurement of the instru-ments used (e.g., dichotomous "yes" = 1 and "no" = 0; continuous or quantitativemeasures) The type of correlation coefficient, be it the intraclass correlation,

Pearson's r, Kendall's g, or Spearman's r, among others, is almost always specified

and justified in research reports.3

The Adoption Study A common way to estimate h 2B is to compare childrenwho are adopted at birth and raised by genetically unrelated individuals The

Variances are computed as the average of the sum of the squared differences between a person's score from the population or sample average.

In this book, I symbolize all correlations (unless otherwise specified) as r for simplicity.

THE EFFECTS OF GENES AND THE ENVIRONMENT 21

comparison of the adopted children with their biological parents provides the

ba-sis for an estimate of h 2B because any measured similarity can only be based on netic similarity (parents and offspring share 50% of the same genetic material onaverage) Environmental factors cannot contribute to the similarity of theadopted children and their biological parents because members of an unrelatedfamily raised the children

ge-In this design, estimates of h 2B are compromised by three factors The first isthat in many modern adoptions the birth mother often maintains some contactwith her child and the adoptive family The degree of contact confounds any esti-mate of genetic similarity with environmental factors A second factor is selectiveplacement of adoptees in families with similar backgrounds For example, it is thepolicy of many adoption agencies to try to place children of a particular ethnicityinto families of the same ethnicity In this situation, the potential of the birth par-ents sharing genes with the adoptive family is increased The risk is higher thanone may think Often, people of the same ethnicity can trace their ancestries back

to common cities, towns, or counties in their ancestral homelands and find thattheir families have intermarried for generations Third, the ability of researchers

to locate sufficient numbers of birth parents to compute the birth spring comparisons is a concern If birth parents are not available, heritability es-timates can still be computed by comparing siblings who have been placed indifferent adoptive homes These families are now quite rare because many juris-dictions have policies to ensure that all children from the same family are adoptedinto a single home

parent-off-The Twin Study A straightforward way to estimate h 2B is to compare the larity of identical or monozygotic (MZ) twin pairs who were separated at birth andraised apart Any similarities between raised-apart MZ twins can only be due totheir shared genes Moreover, unlike adoptee siblings, twins are matched for age,which effectively removes any generational differences as a possible source of error.4

simi-In this design, the computation of h 2B is accomplished by correlating scores on thebehavioral measure between the siblings in each pair If the correlation is 40, thiswould mean that 40% of the individual differences on this measure are due to ge-netic differences between the twin pairs

Finding sufficient numbers of MZ twins raised apart is difficult, so studies oftwins reared together have become the most popular behavioral genetic study de-sign, mainly because of ease of recruitment: Twins who were raised together are rel-atively plentiful (most national censuses and birth record surveys suggest that twinsmake up about 2% of the total population) and they tend to keep in touch with one

another Estimating h 2B on behavior is accomplished by comparing the similarity of

MZ twins on the behavior of interest to the similarity of fraternal or dizygotic (DZ)

It has been argued that being the first child out of the womb may have some kind of beneficial pact, rendering age differences important However, there has been little empirical research published to date showing that the effect is significant or important to the development of psychological problems.

im-22 2 THE ABCs

twin pairs Genetic effects are suggested if the similarity of MZ pairs is found to be

greater than that of DZ pairs Figure 2.3 demonstrates how h 2B is estimated usingdata from twins who were reared together

Estimating the Influence of the Environment

Shared Family Environment Returning to Fig 2.2, the term E represents all

sources of environmental influence on observed individual differences in havior Environmental influence is subdivided into two broad classes The first

be-is the shared or common environment, symbolized by the statbe-istic c2, where the c

refers to the environment (events, conditions, and experiences) that is common

to all members in a household These effects include virtually anything thatcauses members from the same family to become more similar to one another.Like the heritability estimate, c2 is a variance that indexes the proportion of thevariability in behavior due to differences in family environment between house-

holds in a population A frequently used example of c 2 is total family incomemeasured in dollars The degree to which a family is above or below the povertyline affects each person within the family in the same way, but differentiates be-tween families in a sample

The presence of c 2 is readily ascertained in most behavioral genetic designs.For example, in the adoption or twins-reared-apart design the correlation of

adoptees with their adoptive parents yields an estimate of the influence of c 2

Any similarity between the adopted child and adoptive parents can only be due

to sharing the same family environment Computation is a little more

compli-Twin Correlations The source of the similarity between twins

Step 1 Twin Similarities

r MZ = 43 100% genes + 100% common family environment

rDZ = 31 50% genes + 100% common family environment

Step 2 Comparing Twin Similarities

T MZ ~ T DZ = -12 50% genes + 0% common family environment

Step 3 Estimating All of the Genetic Influence

2 (rMZ - rDZ) = 24 2 x 50% = 100% genes estimated

Step 4 Convening to a Proportion

h 2B = 24 x100% = 24%

24% of the individual differences in behavior X is due to all of the genetic

differences between people

FIG 2.3 Estimating heritability

THE EFFECTS OF GENES AND THE ENVIRONMENT 23

cated in the twins-reared-together design Here, the MZ correlation is able to the fact that both members of each pair share 100% of their genetic

attribut-endowment and grew up in the same home In order to estimate c 2 , h 2B must beestimated first and then subtracted from the MZ correlation, leaving an esti-mate of the proportion of the variance due to common family environment Fig-

ure 2.4 illustrates how c 2 is estimated

Nonshared Environment Although the similarity of MZ twins with respect to

major physical characteristics like height or weight is quite high, the correlations arenot perfect MZ twins raised together may share all of their genes and grow up in thesame home, but the correlations between them have been shown to fall well below1.0 What factor accounts for measured differences between identical twins raisedtogether? Assuming that the home environment is the same for both, each musthave had unique experiences This kind of environmental effect is called the

nonshared environment, symbolized as e 2

Nonshared environmental influences are defined as any experience, milieu, orcircumstance that causes children from the same family to be dissimilar It would beincorrect to characterize nonshared environmental influences as solely random(e.g., one twin is involved in a motor vehicle accident and the other is not); they canalso be experiences that systematically differentiate people (e.g., parents systemati-

cally favor one over the other) Figure 2.5 illustrates the computation of e 2 in areared-together twin design From this example, the nonshared environmental in-fluence is what remains after genetic similarity and common environmental factorsare removed It is important to remember that estimates of e2 may be inflated by er-rors of measurement If, for example, a clinical interview on one twin was done nor-mally, whereas the interview for the other was rushed, measurable differencesmight translate into inflated estimates of e2

Estimation of the magnitude of shared environmental effects c 2 using Falconer's method

If h 2B = 24% (from Fig 2.3) and rMZ = 43 then

24 2 THEABCs Estimation of the magnitude of nonshared environmental effects e 2 using Falconer's method.

If h 2B = 24% (from Fig 2.3) and r MZ = 43 then

FIG 2.5 Estimation of the magnitude of nonshared environmental effects e 2 using Falconer's method.

ASSUMPTIONS UNDERLYING THE TWIN METHOD

The Assumption of Equal Environments

The validity of the twin method rests on the assumption that the family ment in which MZ twins are raised is not qualitatively different from the family en-

environ-vironment of DZ twins This is called the assumption of equal enenviron-vironments (EEA).

As shown in previous sections, the estimates of h 2B , c2, and e 2 are predicated on levels

of genetic similarity between relatives Estimates of genetic effect would be cially increased if it could be shown that the similarity of MZ twins is not simply be-cause they share 100% of their genes, but also because they are treated moresimilarly than DZ twins MZ or DZ twin similarities beyond levels of genetic simi-larity can be caused by family members, friends, school teachers, and so on whotreat twins differently from nontwins Moreover, the similarity of MZ twins may beartificially inflated because identical twins are dressed alike more often or treatedmore similarly by their parents than are DZ pairs

artifi-The EEA in most twin studies can be tested by asking twins of all zygosity groups(e.g., MZ male, MZ female, DZ male, DZ female, and DZ opposite-sex pairs) aboutthe degree to which they were treated alike, dressed alike, placed in the same class-rooms, and so on A typical list of these questions is presented in Fig 2.6 A statisticallysignificant difference on any of these items would suggest that the EEA was not sup-ported and that any estimates of genetic and environmental effect may be biased.Fortunately, research has shown that violations of the EEA have had only a mi-nor impact on estimates of genetic and environmental effect For example,Borkenau, Reimann, Angleitner, and Spinath (2002) found that, although MZtwins reported being treated more similarly than DZ twins, this did not translateinto increases in measured similarity in personality Similarly, Kendler, Neale,Kessler, and Heath (1994) found that violations of the EEA had a minor correctableimpact on heritability estimates for several psychiatric conditions It is importantfor readers of behavioral genetic research to be aware of how a published study tests

ASSUMPTIONS UNDERLYING THE TWIN METHOD 25

We spend most of our time together True or False

We attend the same school True or False

We have the same friends True or False

We tend to dress alike True or False

We are in most of the same classes at school True or False

We have always spent a lot of time together True or False Our parents treated us pretty much the same True or False

We have never been apart for more than 1 month True or False

We have almost always had the same teachers True or False

We try to be different from one another True or False Have you ever been separated from your twin for more than 1 month before the age of 18? If yes, please indicate where and with whom each of you lived, what you were doing, the reason for the separation, and your age at the time.

Have you had any important experiences or training that your twin has not had? Please explain.

FIG 2.6 Sample questions used to assess the validity of the equal environments assumption.

whether or not the EEA holds If a violation is reported, it is important to determinewhether the reported estimates of heritability have been adjusted for its effect

Zygosity Diagnosis

Another threat to the validity of the heritability estimate is inaccuracy in zygositydiagnosis For example, MZ pairs misdiagnosed as DZ spuriously inflate the DZcorrelation while simultaneously decreasing the MZ correlation Heritability anal-

yses underestimate h 2B and overestimate c 2 The same is true when DZ pairs are

misdiagnosed as MZ

The best way to determine zygosity is to compare the similarity of DNA morphisms extracted from blood or buccal cells However, the expense of DNAanalysis is prohibitive Many studies instead rely on questionnaires to assesszygosity These questionnaires make their diagnoses based on twins' degree ofphysical similarity and the extent to which people of differing degrees of acquain-tance confuse them Figure 2.7 presents examples of these items Their validity hasbeen demonstrated by several studies that have shown questionnaires to be accurate

poly-at least 94% of the time compared to DNA analysis (e.g., Kasriel & Eaves, 1976).The problem of zygosity misdiagnosis does not appear to concern many re-searchers because, if anything, they are left with a statistically conservative estimate

of heritability It is preferable to miss a possible genetic effect than to say that one ispresent when it is not The latter could lead to very expensive genotyping studiesthat are destined to fail

As you know, there are two kinds of twins: identical (or monozygotic) twins who have the same heredity, and fraternal (or dizygotic) twins, who only share part of the same heredity The following questions are intended to help determine which kind you are.

1 What is the natural color of your hair? If your hair is different from that of your twin in any of the following ways, please describe these differences:

2 What is the color of your eyes?

3 How tall are you? How much taller or shorter are you than your twin?

4 How much do you weigh? How much heavier or lighter are you than your twin?

5 If you know your blood type and Rh factor, please indicate them here.

6 As a young child, did your parents ever mistake you for your twin?

11 Do you and your twin look alike? Please explain.

12 If you know whether you are fraternal or identical, how do you know? How and by whom was it determined?

FIG 2.7 Examples of questions used to diagnose zygosity.

26

environ-To estimate them, model-fitting approaches must be used.

A modelis an idea that has been formalized into a diagram This diagram, called a

path diagram, explicitly describes the hypothesized relationship between variables

(e.g., A causes B) using the standard drawing methods of path analysis (e.g., A —> B).

In this way, entire systems describing how variables affect and are affected by eachother can be drawn The arrows are converted to mathematical equations like thoseused to compute regression and correlation coefficients Data are collected on allthe variables that are then run through these equations to determine which of thehypothesized relationships are supported (e.g., the value of the arrow, also called a

parameter, in A —> B is greater than zero) If the data fit the model, the model is

sup-ported If the data do not fit the model, the model is revised and the process repeated

(e.g., trying B —> A) This process is called model fitting.

Behavioral geneticists have embraced model fitting because it is the only

method that can separate additive and dominance genetic effects from h 2B , as well

as estimate c 2 and e 2 at the same time, as shown in the path diagram in Fig 2.8 Thesquare boxes in Fig 2.8 represent the actual scores (e.g., scale scores from a self-re-port inventory or diagnosis) for each member of a twin pair The circles representwhat we cannot directly observe, in this example, the genetic and environmentalfactors hypothesized to influence the variability of the measured variables: addi-tive genetic (A), genetic dominance (D), shared environmental effects (C), and

nonshared environmental effects (E) The straight one-headed arrows or paths from the circles to the squares represent the hypothesized influence of A, D, C, and E on each of the observed variables These paths are labeled a, d, c, and e, re-

spectively The values of the paths represent the strength of the relationship

be-tween A, D, C, and Eon each of the observed variables and are used to compute the heritability estimates, h 2A , h 2d , c 2 , and e 2 , respectively.

The double-headed curved arrows represent the hypothesized relationship tween the circles, and the values differ for MZ and DZ twins Starting with MZtwins, the curved arrow between the additive genetic effects for each twin (between

be-A Twin1 and A T w i n 2 ) would be set at 1.0 because they share 100% of their genetic

ma-terial For the same reason, the paths between the genetic dominance influences(DTWIN and -DTWIN 2) will also be set at 1.0 The path between the shared environ-mental effects (C TWIN 1 , and CTWIN 2) is also set at 1.0 because, in a reared-togethertwin design, both members grow up in the same home There is no path between the

sources of nonshared environmental effects (E TWIN l and E TWIN 2) because, by tion, these effects are unique for each member of a pair